VoiceBots (previously known as Cognitive IVR) uses Google Cloud Speech-to-Text to improve the performance of natural-language interfaces such as Dialog Engine. This makes it easier for callers to use spoken natural-language phrases to navigate through an Genesys Intelligent Automation application.
For example, a traditional IVR might have trouble interpreting the following phrase: "I would like to book a flight to Paris on Friday afternoon." Instead, the traditional IVR would need to ask several sequential questions to deduce the following information:
- What do you want to do? (Book a flight)
- Where do you want to travel? (Paris)
- When do you want to travel? (Friday afternoon)
However, VoiceBots can interpret and parse natural language so the customer can make the same one-sentence request without having to sequentially answer several questions: "I would like to book a flight to Paris on Friday afternoon."
Other than a new login screen, using VoiceBots does not affect how you use Genesys Intelligent Automation. You can build and use applications in the same way (but note the limitations below) and share the same database.
Integrate Intelligent Automation with Google Speech-to-Text
When setting up Intelligent Automation for use with voice, a start page is required. For VoiceBots that use speech transcription, the start pages would be:
- GenesysGVP8_GoogleSR_Vocalizer6.jsp if you're using GVP 8 and 9
- Interactive_GoogleSR_Vocalizer5.jsp if you're using PureConnect 2018R4 and above.
The start page should contain the following parameters:
- testsiteid - The ID of the application being provisioned.
- istestcall - When set to True, the provisioning uses the test mode (the latest saved version) or the production mode (the latest deployed version) when set to False.
- authtoken - This will be available from the Company Details page.
- mrcpversion - This is an optional parameter. The default value is 1.
An example start page that uses GVP: http://<server>:<port>/fish-vui/start/GenesysGVP8_GoogleSR_Vocalizer6.jsp?testsiteid=53&istestcall=true&authtoken=303a935e028b1aae234476fed4a591c3b21601b40d3d1351e23384bc0b99700d&mrcpversion=2
The RTP is streamed to Google STT and the resulting transcription is sent to Dialog Engine or DialogFlow for processing.
The following limitations apply to the use of VoiceBots:
- Multimodal communication is not supported.
- The Release ASR API command for the Script block does not have any effect when using VoiceBots.
- WebIVR applications are not supported.
- For a given persona, the TTS language is also used for the speech-recognition language.
You need the following to use VoiceBots with Genesys Intelligent Automation: